WHAT IS DONE
· Developed a Design Research framework to understand the effect of social network’s features and the polarity of the provided content on information flow and collective opinion about a dichotomized topic (e.g. politics) on social media
· The framework is a two-step approach; simulation and machine learning.
o Implemented simulation on MATLAB to simulate the information flow on the network and measure the evolution of collect opinion over time
o Leveraged machine learning to interpret the simulation results and unravel potential mechanisms of the collective opinion formation
With the immersion of social media in our lives, not only the way we interact with other people, but also the way our opinions are formed changed. In this study, information sharing and its effects collective opinion changes in digital context is being investigated. As digital context, Twitter platform is selected since it is a microblogging website and it has asymmetric connections, which makes network structure directed and more generic. In order to predict collective opinion formation, a framework has been proposed which is a combination of Monte-Carlo simulation and Machine Learning. Based on Statistical Mechanics Theory, macro-level outcomes of complex systems could be calculated despite high levels of freedom. Monte Carlo method has been used to analyze complex systems, and the designed artifact uses Monte Carlo Simulation to feed Machine Learning algorithm so that emergent behaviors on the network will be predicted with the help of the designed framework. Preliminary runs show that simulation exhibits emergent behavior. In the future, the proposed framework will be further analyzed in terms of efficacy and validity, and comparative case studies will be made to analyze utility of the framework.
Keywords: Collective Opinion, Opinion Dynamics, Design Science Research, Monte-Carlo simulation, Machine Learning.